U.S.-based Oracle (ORCL) announced on July 14, 2026, the expansion of “Oracle AI Agent Studio for Fusion Applications,” a new development platform that fully operationalizes AI agents within its integrated cloud ERP, “Oracle Fusion Cloud Applications.” The platform provides a seamless continuum from no-code development using natural language to pro-code development utilizing Codex and Claude Code, deploying 22 agentic applications for finance, human resources, and supply chain. Enterprise systems are evolving from “tools for operating screens” into “platforms where AI executes business tasks under governance.”<\/p>\n
This announcement represents more than just the addition of AI features. Oracle’s Chris Leone emphasized that this approach is “fundamentally different” from the conventional method of building AI externally and then adding governance functions afterward. To date, many companies have begun AI adoption by placing chatbots or search functions alongside their ERP and CRM systems. However, entrusting AI with tasks involving permissions and approvals\u2014such as invoice processing, purchase requisitions, and HR procedures\u2014has been limited by these bolt-on mechanisms. Oracle’s new platform solves this challenge by running AI directly on top of Fusion Applications’ existing business data, workflows, approval flows, security, and audit functions.<\/p>\n
The newly offered “Agentic Applications Builder” allows business users to construct applications combining agents, business processes, and reference data simply by providing instructions in natural language. For developers, “AI Studio Skill” provides Visual Studio Code, Git, and CLI (Command Line Interface), enabling the creation, editing, validation, and debugging of agentic applications in a local environment. According to Oracle’s materials, developers can also instruct Codex or Claude Code to perform tasks using natural language.<\/p>\n
The most significant change this approach brings is the restructuring of the division of labor among business departments, developers, and IT departments. Traditionally, these functions operated in silos: business departments communicated requirements, developers built systems, and IT departments conducted security reviews. The new platform enables all three parties to work in parallel while sharing the same applications and business rules. The structure shifts to one where business departments define objectives and exception conditions, developers implement complex processing and external integrations, and IT departments centrally manage permissions, approvals, and production deployments. This represents a transition where the barrier between natural language and code is lowered, embedding business knowledge and governance rules from the earliest stages of development.<\/p>\n
Oracle’s Steve Miranda described a vision of evolving enterprise software from passive “systems of record” into applications that “reason, judge, and act” toward objectives. In a real-world business scenario, for example, in accounts receivable collections, the system can not only identify past-due invoices but also analyze customer payment history, outstanding amounts, recent communications, and contract terms to determine priorities, prepare notifications for responsible staff and customer communications, and then delegate only exceptional judgments to humans.<\/p>\n
Currently, 22 types of Fusion Agentic Applications are available for Oracle Fusion Cloud ERP, HCM, SCM, and CX. In the finance domain, they evaluate overdue balances, invoice aging, and customer responses. In HR, they continuously monitor staffing, leave requests, and work rule issues. In the supply chain domain, they link design changes with procurement, supporting tasks up to sending requests for quotes to suppliers.<\/p>\n
However, running AI agents within an ERP does not automatically guarantee safety. For enterprises to use them in production operations, at least four governance conditions must be designed. First is segregation of duties: the information AI can reference and the information it can update must be strictly separated, preventing situations where AI alone can execute processes not permitted to the user themselves. Second is the approval process: for tasks like purchase orders, payments, hiring, and contract changes, human confirmation must be mandatory based on amount and risk, with mechanisms to halt the AI workflow until approval results are returned.<\/p>\n
Third is the clarification of accountability. When multiple agents execute cross-departmental processes, it is necessary to predefine who among the requester, approver, developer, and business owner is responsible for explaining results, and who holds the authority for exception judgments and operational shutdowns. Fourth is change management: AI agents are not a one-time build. Changes to the model used, instructions, reference data, business rules, and approval paths can all impact behavior. Managing development, validation, approval, production deployment, monitoring, and rollback as a consistent flow is essential.<\/p>\n
Meanwhile, investor concern is growing over Oracle’s massive investments and market risk. Oracle shares fell 6.5% on July 13 and another 2.7% on July 14, once again testing the stock’s downside resilience during a broader market correction. While the company secured $67 billion (approximately \u00a510.9 trillion) in new AI infrastructure contracts in a single quarter, it also projects net cash outlays of approximately $70 billion (approximately \u00a511.4 trillion) for fiscal 2027, with a financing plan of roughly $40 billion (approximately \u00a56.5 trillion) also underway.<\/p>\n
Oracle’s stock performance during past market turmoil offers clear lessons for investors. The company has experienced 15 market disruptions, with an average decline of 18%, exceeding the S&P 500’s 16%. The most severe was a 40% drop during the 2008-2009 global financial crisis, with a 32% decline during the 2025 market turmoil and a 30% fall during the 2022 bear market. Notably, the median recovery time is about four months, but after the 2014-2016 oil price crash, it took nearly 27 months to return to previous levels.<\/p>\n
That said, Oracle today possesses a different business structure than in the past. Its remaining performance obligation (RPO) has reached $638 billion (approximately \u00a5103.7 trillion), and its unique position in AI\u2014providing everything from Oracle Cloud Infrastructure to various applications in an integrated manner\u2014underpins a high degree of certainty for future revenue. However, risks remain, including intensifying competition among major tech companies in the AI data center space. If an investor had allocated 10% of their portfolio to Oracle stock, the largest historical 40% decline would have resulted in an overall portfolio erosion of roughly 4%, once again highlighting the importance of position management and diversification.<\/p>\n
Oracle’s latest announcement marks a shift from the phase of obtaining convenient answers via bolt-on AI to redefining the ERP itself as a platform where AI executes work according to business rules. What will determine corporate competitiveness is moving beyond just the intelligence of AI models to the ability to build mechanisms that can entrust tasks to AI to the greatest extent possible, with embedded permissions, approvals, and accountability.<\/p>\n","protected":false},"excerpt":{"rendered":"U.S.-based Oracle (ORCL) announced on July 14, 2026, the expansion of “Oracle AI Agent Studio for Fusion Applications,”…\n","protected":false},"author":2,"featured_media":111453,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[6],"tags":[405,7537,57260,2798,2317,54878,2163,57259,54876,9848,57261],"class_list":["post-111452","post","type-post","status-publish","format-standard","has-post-thumbnail","category-agentic-ai","tag-ai-agents","tag-artificial-intelligence-agents","tag-chris-leone","tag-claude-code","tag-codex","tag-fusion-agentic-applications","tag-oracle","tag-oracle-ai-agent-studio-for-fusion-applications","tag-oracle-fusion-cloud-applications","tag-orcl","tag-steve-miranda"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/111452","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=111452"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/111452\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/111453"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=111452"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=111452"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=111452"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}